I might add, in the way that a programmer might, that living unformalized theories are very flexible, and trying to "disprove" based on fact will never end the

I might add, in the way that a programmer might, that living unformalized theories are very flexible, and trying to "disprove" based on fact will never end the dispute. It's better to think in terms of "what model responds with fewer epicycle" as new evidence arises, or even broader and more simply "which abstractions are more useful". So it's more like the quest for better programming languages and libraries, than "right" and "wrong". I see no other way to resolve these sorts of disputes productively.